(Mis)Representing Ethnicity in UK Government Statistics and Its Implications for Violence Inequalities

Round 1

Reviewer 1 Report

Hello,

this is a very well presented, well argued and well researched paper. The discussion of classification of ethnicity is very important as the implications of how ethnicity and the effects of crime and fear of crime are experienced are mediated through such classification. Using a practical application also worked well.

Sections 6.1 and 6.2 - this is just a suggestion but due to the volume of data comparisons presented it is difficult to follow the discussion, reconsider the presentation - maybe simple data tables or discussion in bullet point would make these sections more accessible. 

There are a few rogue words/capitalisation in section 7.1 and 7.2 and there is a bracket (need to add) remaining in the text - line 766. These are the only issues with the text spotted. 

The discussion is thorough and well considered and importantly you acknowledge the potential implications of splitting the minority ethnic sample further. There may also be other groups that feel that separate considerations are important (Eastern Europeans, for example) so a nod to how far this breakdown of categories could/should go, may be useful. Finally, you mention that very different experiences of crime are currently grouped into 'other' or 'mixed' but that is also true with the 'white' category - e.g. Gypsy/Traveller as well as other groups so again, it may be worth acknowledging this briefly.

Author Response

Overall Response: We would like to thank you for your supportive comments that “this is a very well presented, well argued and well researched paper”, kind acknowledgement of the importance of this topic, and your comments that “using a practical application also worked well” and that “the discussion is thorough and well considered”. We have made changes to address the clarity of our results presentation and discussion, typos notes, and acknowledgment of additional groups which may be subject to misrepresentation.

R1: “Sections 6.1 and 6.2 - this is just a suggestion but due to the volume of data comparisons presented it is difficult to follow the discussion, reconsider the presentation - maybe simple data tables or discussion in bullet point would make these sections more accessible.”

Response: We would like to thank you for highlighting that the presentation and discussion of the results were difficult to follow. We have now simplified the regression tables in the main text so that the results are based on ‘White’ being the reference category, and instead included the full tables (with rotating referent groups) in appendix D (page 30). We have also amended section 6.1 to simplify and clarify the results discussion (see highlighted lines 653, 657, 662-664, 666, 668-685, 721-728, 731-733), which we agree were more difficult to follow in the previous version of the paper.

R1: “There are a few rogue words/capitalisation in section 7.1 and 7.2 and there is a bracket (need to add) remaining in the text - line 766. These are the only issues with the text spotted.”

Response: Thank you very much for flagging these. We have now corrected these typos and mistakes in the text (lines 816, 818, 882).

R1: “The discussion is thorough and well considered and importantly you acknowledge the potential implications of splitting the minority ethnic sample further. There may also be other groups that feel that separate considerations are important (Eastern Europeans, for example) so a nod to how far this breakdown of categories could/should go, may be useful. Finally, you mention that very different experiences of crime are currently grouped into 'other' or 'mixed' but that is also true with the 'white' category - e.g. Gypsy/Traveller as well as other groups so again, it may be worth acknowledging this briefly.”

Response: We appreciate your supportive comments on our discussion and acknowledgment of the potential risks of further splitting the minoritised ethnic sample, and take your point on additionally acknowledging groups which we did not include in our alternative categorisation. We have now included this on page 21 (lines 840-846).

Additional Amendments:

We would also like to highlight an addition to the paper which sought to address potential issues of reliability of results when using a mixed sample of migrants and non-migrants. That is, that due to data availability and limitations when recategorizing ethnic groups, most ethnic groups included migrants and UK-born respondents whilst the ‘Latinx/Hispanic’, ‘ESEC Asian’ and (pre-2011) ‘Arab/MENA’ groups were migrant-only. We re-ran our analyses with a migrant-only sample to enable a like-for-like comparison across the new groups, and the results supported our overall results and conclusions in our main analyses. This is briefly noted in the methodology (page 9, lines 373-375 and footnote 4) and results (page 16, lines 642-645), with supplementary results tables included in appendix E.

Additionally, we have updated the results in table 2 (page 10), since these now use the analytical sample, excluding cases with missing values, and not on the full sample. This update involves only minor superficial changes and does not change any of our findings. The updated results are reflected in section 5 (page 9, lines 386-8 and lines 392-6; page 12, lines 432-433 and 464; and page 13, line 496).

We also added the point on re-introducing the ethnic-boost sample to the recommendations section (page 22, lines 952-5), which was noted earlier in the paper but not explicitly mentioned in the recommendations.

We also corrected our funding statement with further details and additional acknowledgment (page 24, lines 1043-1053) and author contributions (page 24, lines 1037-1041)

Finally, we have made minor typo and wording corrections on page 7 (footnote 3: ‘Appendixes 1 & 2’ corrected to ‘Appendices A & B), page 12 (line 462, 478), page 13 (lines 481, 486-7, 491, 494, 501-2), page 14 (lines 550, 556-8), page 15 (lines 617-8), page 19 (lines 774, 785, 805-7, 816, 818), page 20 (lines 823, 825, 839, 848-863 [to avoid repetition]), page 21 (883, 899, 912-3), page 22 (lines 934, 938), page 23 (989, 1005, 1022-4), and appendix A. Corrections are made via tracked changes and highlighted in yellow.

Reviewer 2 Report

The article presents the problem of measuring ethnicity in ethnically diverse societies. It assesses the challenges, inconsistencies and limitations of the current approach to measuring ethnicity through a critical evaluation of standardised ethnicity measures in UK survey data (particularly the CSEW) and analyses CSEW data to compare different approaches to measuring ethnicity and its implications for violence-related outcomes. It brings useful methodological recommendations.

Suggestions for improvement:

Table 1 - title should be shortended - there is not enough space for the current title

Table 2 - typo in heading (orgin -> origin)

7.1 and 7.2 - redundant word "Using" in first paragraphs

Appendix C - formatting should be improved and regions separated by spaces or lines

Author Response

Overall response: We greatly appreciate your careful consideration and comments on the contributions of this manuscript, particularly in “bring[ing] useful methodological recommendations”. With thanks for very helpfully flagging these, we have made improvements to formatting and corrected typos.

R2: “Table 1 - title should be shortened - there is not enough space for the current title”

Response: We have now shortened the title so that it fits on the page correctly (page 8)

R2: “Table 2 - typo in heading (orgin -> origin)”

“7.1 and 7.2 - redundant word "Using" in first paragraphs”

Response: Thank you, we have now corrected these typos: removed ‘using’ from lines 816 and 883; ‘origin’ typo corrected in table 2 (page 10).

R2: “Appendix C - formatting should be improved and regions separated by spaces or lines”

Response: We have sought to improve the formatting and separated regions (first column) by lines so they do not wrap, improving the presentation of this table.

We would also like to highlight an addition to the paper which sought to address potential issues of reliability of results when using a mixed sample of migrants and non-migrants. That is, that due to data availability and limitations when recategorizing ethnic groups, most ethnic groups included migrants and UK-born respondents whilst the ‘Latinx/Hispanic’, ‘ESEC Asian’ and (pre-2011) ‘Arab/MENA’ groups were migrant-only. We re-ran our analyses with a migrant-only sample to enable a like-for-like comparison across the new groups, and the results supported our overall results and conclusions in our main analyses. This is briefly noted in the methodology (page 9, lines 373-375 and footnote 4) and results (page 16, lines 642-645), with supplementary results tables included in appendix E.

Additionally, we have updated the results in table 2 (page 10), since these now use the analytical sample, excluding cases with missing values, and not on the full sample. This update involves only minor superficial changes and does not change any of our findings. The updated results are reflected in section 5 (page 9, lines 386-8 and lines 392-6; page 12, lines 432-433 and 464; and page 13, line 496).

We also added the point on re-introducing the ethnic-boost sample to the recommendations section (page 22, lines 952-5), which was noted earlier in the paper but not explicitly mentioned in the recommendations.

We also corrected our funding statement with further details and additional acknowledgment (page 24, lines 1043-1053) and author contributions (page 24, lines 1037-1041)

Finally, we have made minor typo and wording corrections on page 7 (footnote 3: ‘Appendixes 1 & 2’ corrected to ‘Appendices A & B), page 12 (line 462, 478), page 13 (lines 481, 486-7, 491, 494, 501-2), page 14 (lines 550, 556-8), page 15 (lines 617-8), page 19 (lines 774, 785, 805-7, 816, 818), page 20 (lines 823, 825, 839, 848-863 [to avoid repetition]), page 21 (883, 899, 912-3), page 22 (lines 934, 938), page 23 (989, 1005, 1022-4), and appendix A. Corrections are made via tracked changes and highlighted in yellow.